The IEEE 802.3af Power over Ethernet (PoE) standard was ratified last year. This standard defined the protocols for safely distributing 48-Vdc power over standard Ethernet cabling (typically Cat 5) from networking equipment to any network peripheral designed to accept this power. In the language of the standard, the networking equipment (Ethernet hub, switch or router) became the power sourcing equipment (PSE), while the network peripheral became the powered device (PD) (see the figure).
To distinguish a PD from a network peripheral that could not accept PoE power, the PD would present a 25-kΩ signature resistance when plugged into the Ethernet port. The PSE would detect this resistance before it attempted to apply 48-V power to the Ethernet port. Optionally, the PD could present another signature for power classification.
The PSE also would need to limit inrush current during start up to the PD and detect when a PD has been removed from a port. Power is transmitted using either the spare pairs of the Cat 5 cable in the case of a mid-span device (a piece of equipment that retrofits existing LAN equipment for PoE) or over the signal pairs in the case of an end span device (where PoE function is incorporated into new LAN equipment).
IEEE 802.3af established a continuous current limit of 350 mA per Ethernet port. Accounting for losses in the cabling, this permitted each PD to draw up to 12.95 W max. That limit was more than enough to power most VoIP phones and sufficient to power many other types of devices that might be connected to the LAN. Aside from VoIP phones, the two most-mentioned applications are WLAN access points and security cameras.
A number of semiconductor vendors were quick to develop silicon to integrate these PoE functions so that PSE and PD circuitry could be implemented at lowest cost with the minimum number of components. Power Dsine, Texas Instruments, Maxim Integrated Products, Linear Technology and Supertex were among the first chip makers to introduce what would be known as PSE and PD power controllers (or power managers). These devices address the requirements set forth in IEEE 802.3af. However, they also offer some provisions for equipment makers who must ensure compatibility with the legacy (proprietary) PoE systems that predate the IEEE standard.
Companies that develop PoE products can verify their performance and their compliance with the standard by participating in multi-vendor tests conducted by the University of New Hampshire's InterOperability Laboratory (UNH-IOL). Chip makers who develop PoE controllers and reference designs are among the companies participating in these tests. The UNH-IOL's PoE Group Test Period allows vendors to test their PoE equipment in a small, neutral setting. They can then correct any problems discovered during testing, rather than waiting for these problems to show up in the field.
Although standardization defines a basic level of functionality that must be met, certain performance and implementation factors (standards conformance, reliability, packaging, level of integration, required board space and system cost) distinguish one vendor's offerings from another.
Going forward, pressure to integrate more of the PSE and PD control functionality will continue. To some extent, this will mean clever partitioning of PoE control functions and PHY functions to get the most advantageous implementations. It will also involve the integration of power conversion circuitry and PoE functions, particularly for the PDs. National Semiconductor, Fairchild Semiconductor, Micrel Semiconductor, Semtech and Intersil are among the other IC companies looking to leverage their power design expertise to optimize PoE designs.
For more information on this article, CIRCLE 342 on Reader Service Card